quanteval.ai tracks how quantized open-source models perform on real coding, reasoning, and tool-use work. This page highlights the strongest quant from each family in each benchmark suite.